Abstract: After SA was terminated,the GPS positioning accuracy using C/A code recovered to above 15m (DOP=2.5).And USA is planning to apply some innovations to the new generation of GPS,thus the SPS accuracy of code positioning will be better improved in the future.Since GPS is under the control of American government,we have to face the problem how to make use of the superior navigation performance of GPS in our military applications and to avoid the potential dangers as well.China has founded it's own RDSS system.According to the tests in Japan and USA,we are confident that the differential positioning accuracy of our RDSS will not be worse than 50m (2D).But RDSS still has the problem of low sampling rate,and what's more,the central processing station adopts a classification response mechanism,so the response delay of a low-grade user may be much longer than 1s.Therefore it's difficult for us depending only on RDSS to meet the demands of high dynamic users.Another deficiency of RDSS lies in the low reliability,because the space segment of RDSS consists of only two satellites.To fulfill the requirements of the high-dynamic applications and provide high reliability,this dissertation firstly proposed aiding RDSS with INS to develop the INS/RDSS integrated navigation system.Further more,in order to make full use of GPS,considering the GPS equipments already in use,we can also integrate GPS into the INS/RDSS integration.The new multi-sensor integration of INS/RDSS/GPS will not only achieve the high accuracy of the INS/GPS integration,but also increase the system reliability,which is the essential demand of the most navigation and guidance systems.So it will be of great significance developing this INS/RDSS/GPS integration,especially in military.Multi-sensor data processing methods can be classified into two basic types:centralized filters (CKF) and distributed filters.In practice,CKF may result in severe calculation burden,and the failure of any sensor will contaminate the whole system.To deal with the limitations of CKF,the distributed (or decentralized) filtering technology has received increasing attention.Speyer and Willsky made early contributions to the distributed filter theory,and then Kerr proposed a filtering structure with attractive fault-tolerant feature,but he failed to present a practical realization.Recently,Carlson developed the federated Kalman filter (FKF) method,which provides globally or near-optimal estimation accuracy based on information sharing principles.FKF realized the filtering structure of Kerr.It can be classified into two design types or four configuration modes in order to meet different design and capability demands.In these four modes,the NR mode shows a high degree of fault-tolerance and strong fault detection capability.This paper shortly presents the distributed federated filter theory put forward by Carlson,and discusses the two design types and four configuration modes.The idea of developing the SINS/RDSS/GPS integration is put forward for the first time;the characteristic of each local system is discussed and then the mathematic model of the INS/RDSS local system is constructed.It's proposed in this paper that the NR configuration mode,which belongs to design type A of FKF,should be applied to the practical development of SINS/RDSS/GPS system.The numerical results show the near-optimality and good fault-tolerance and failure-detection performance of the NR mode.